General Skating Chat

[hoodie axel]

On 2/9/2019 at 9:53 PM, Xen said:

I was kind of curious if anyone has ever bothered to plot out the men's and ladies' PCS scores to see the normal distribution and where skaters sit on a bell curve. 

Part of the reason is to figure out what is the avg, mean and actual scores given to skaters, relative to each other, as PCS does seem to factor in skater's relative performance vis-a-vis each other. 

 

So I've gathered the data for the ladies at 4CC, taking out Brooklee Han as she WD in the long. So of the 21 skaters, my current results are:

Component	Mean	Median	St Dev
SS	7.10	6.95	0.88
TR	6.79	6.59	0.97
PE	7.04	6.93	0.99
CO	7.08	6.89	0.93
IN	7.11	7.02	0.99

 

Sheet is here: (link)

I actually don't use excel that often anymore (there's a project software that does all these calculations for me at work), and I'm definitely not a statistician. But I thought having a good normal distribution chart, with a decent indication of where people sit on a bell curve, might give clues as to how big a factor nationality is, and evaluate (vs actual PCS guidelines) where skaters are considered ranked next to each other.  Then maybe we can start actual discussions of whether or not skaters are sitting where they should be sitting on the actual bell curve. 

I do not know if I'd use a bell curve to fit this data. There are stresses on the scores which would change the distribution... (your flowchart for instance)

On 2/25/2019 at 8:10 AM, kaeryth said:

 

Sneak peek with Dai

 

Do we know if they're going to perform a traditional play, or have they also written something new for this purpose? The latter would be even cooler.

[kaeryth]

Opening Ceremony | Winter Universiade 2019

Spoiler

 

 

[hoodie axel]

https://goldenskate.com/2019/03/mona-adolfsen-on-being-an-isu-judge-and-the-new-scoring-system/

 

Quote

Pre-rotated and under-rotated jumps in figure skating are a much discussed topic.

“If it is very clear that the rotation is done on the ice before the take-off, the jump is downgraded and evaluated using the scale of values for the jump of one rotation less,” Adolfsen explained. “The position of the toe pick/skating blade on the take-off is considered here, not the body position as some may believe. The position of the blade where the toe pick hits the ice is considered to assess whether the jump is under-rotated or not.”

“We cannot see the jumps in slow motion, otherwise there will soon be many jumps to be considered and may be downgraded,” she continued. “Again, there are small margins, and here it has been decided that it must be visible in normal speed to consider whether the jump is pre-rotated or not. The rule for pre-rotations is not as strict as for landing of jumps as it is physically impossible to not have any pre-rotation.”

 

But it's not 

[shanshani]

I spent some time looking at GOE scoring in preparation for worlds, specifically Yuzu versus Nathan's raw GOE marks. Here's what I found:

 

(this is all GP circuit-only data)

 

Yuzuru Hanyu average raw GOE: 2.56
Nathan Chen average raw GOE: 1.98
Difference: 0.58

 

I went ahead and looked at what the average GOEs were if we excluded negative GOEs as well (so in other words, how do they score when they don't screw up)

Yuzuru Hanyu average raw GOE (positive only): 3.44
Nathan Chen average raw GOE (positive only): 2.94

I also looked specifically at positive quad jumps and 3As, since the bulk of actual GOE points are earned on those elements. What I found was interesting:

Yuzuru Hanyu average raw GOE (clean quads/3As): 3.54
Nathan Chen average raw GOE (clean quads/3As): 2.48
Difference: 1.06

That's a huge difference! (but imo completely right, Yuzu's jumps *are* at least 1 point better than Nathans ) But what if we look only at top scoring big jumps?

Yuzu:
4T (GPRUS FS) - 4.44
4S (GPRUS SP) - 4.33
4S (GPFIN SP) - 4.33
3A (GPRUS SP) - 4.11
4S (GPRUS FS) - 3.78

average: 4.20

Nathan:
4F (GPF FS) - 3.89
4T (GPFRA FS) - 3.78
3A (GPUSA SP) - 3.67
4F (GPF SP) - 3.44
4T-3T (GPF FS) - 3.33

average: 3.62

difference: 0.58

 

More modest, but the difference is still clear. But moreover, this means my assumptions when I was analyzing a hypothetical Yuzu versus Nathan head-to-head are pretty good, and if anything underestimate Yuzu. The conclusions of that analysis were:

 

1. Yuzu wins handily over Nathan if Yuzu does the Helsinki layout and Nathan does the Nats layout and they're both clean

2. Yuzu squeaks a victory over Nathan (but imo it's pretty much a statistical dead heat) if Nathan adds 4S and they're both clean

 

so nope, Johnny, clean Nathan is not unbeatable.

[hoodie axel]

2 hours ago, shanshani said:

so nope, Johnny, clean Nathan is not unbeatable. 

Does anyone in their right mind believe this anyway? I can make several stastically unlikely assumptions and conclude that Slavik Hayrapetyan will be WC 2019. Hardly a shock that Weir hasn't taken a 5th grade course in mathematics.

[Old Cat Lady]

8 hours ago, shanshani said:

Yuzu:
4T (GPRUS FS) - 4.44
4S (GPRUS SP) - 4.33
4S (GPFIN SP) - 4.33
3A (GPRUS SP) - 4.11
4S (GPRUS FS) - 3.78

Nathan:
4F (GPF FS) - 3.89
4T (GPFRA FS) - 3.78
3A (GPUSA SP) - 3.67
4F (GPF SP) - 3.44
4T-3T (GPF FS) - 3.33

 

I altered the form you made for judging to calculate raw GOE (I have 3 versions of your form now, lol) and checked the factored GOE's against ISU protocols.  I'm not sure if my form is off or yours is but I have a few minor quibbles with your numbers.  I got:

 

Yuzu:
4T (GPRUS FS) - 4.57
4S (GPRUS SP) - 4.43
4S (GPFIN SP) - 4.43
3A (GPRUS SP) - 4.14 
4S (GPRUS FS) - 3.71

Nathan:
4F (GPF FS) - 4
4T (GPFRA FS) - 3.72
3A (GPUSA SP) - 3.71
4F (GPF SP) - 3.43
4T-3T (GPF FS) - 3.43

 

https://docs.google.com/spreadsheets/d/1bz2lk-wF5FlI16_lmgUFfL85MmI_DS7M7AKIjE6fW6E/edit?usp=sharing

(I was using this sheet to see how they would have scored if they didn't make mistakes so some of the GOE's/levels are altered.  Also, I didn't attempt a formula for the choreo sequence so I just estimate and manually enter those)

 

[shanshani]

1 hour ago, Old Cat Lady said:

 

I altered the form you made for judging to calculate raw GOE (I have 3 versions of your form now, lol) and checked the factored GOE's against ISU protocols.  I'm not sure if my form is off or yours is but I have a few minor quibbles with your numbers.  I got:

 

Yuzu:
4T (GPRUS FS) - 4.57
4S (GPRUS SP) - 4.43
4S (GPFIN SP) - 4.43
3A (GPRUS SP) - 4.14 
4S (GPRUS FS) - 3.71

Nathan:
4F (GPF FS) - 4
4T (GPFRA FS) - 3.72
3A (GPUSA SP) - 3.71
4F (GPF SP) - 3.43
4T-3T (GPF FS) - 3.43

 

https://docs.google.com/spreadsheets/d/1bz2lk-wF5FlI16_lmgUFfL85MmI_DS7M7AKIjE6fW6E/edit?usp=sharing

(I was using this sheet to see how they would have scored if they didn't make mistakes so some of the GOE's/levels are altered.  Also, I didn't attempt a formula for the choreo sequence so I just estimate and manually enter those)

 

were you using the trimmed mean? I just used the straight mean of all the judges (as in, I didn't remove the highest and lowest GOE judges from my calculations)

[Old Cat Lady]

51 minutes ago, shanshani said:

were you using the trimmed mean? I just used the straight mean of all the judges (as in, I didn't remove the highest and lowest GOE judges from my calculations)

 

Ahh. That would explain it.  Well, I guess that just means we have some extra data.

 

eta: btw, using the trimmed mean puts it slightly more in Yuzu's favor.  

Yuzu's average = 4.256

Nathan's average = 3.658

Difference = .598

[hoodie axel]

11 hours ago, Old Cat Lady said:

Nathan:

3A (GPUSA SP) - 3.71

[Old Cat Lady]

4 hours ago, hoodie axel said:

 

That was my reaction too.  I was griping about that for weeks.  Tuk stalks half the rink to do a huge, beautiful triple axel and she gets 0-2 GOE (as she should considering the long prep deduction).  Nate stalks half the rink to do a tiny 3 axel and judges throw a bunch of 4's at him 

 

And Nate's fans are still talking about how the Skate America panel was tougher than any of the other ones

[hoodie axel]

Just now, Old Cat Lady said:

And Nate's fans are still talking about how the Skate America panel was tougher than any of the other ones

[shanshani]

Yeah judges seem to be very inconsistent about deducting for long preparation...and rewarding jump size...and entries...and timing with music.

[WinForPooh]

@ICeleste Bringing your video here!

 

The half-assed 'oh yeah must do a tr here' thing done four times is worth all that now. 

 

4 hours ago, ICeleste said:

@WinForPooh your wish is my command! (I've made the video you asked for)

Here are the three World Medallists' free programs, except that I cut out all of their jumps and spins. AKA in this video you'll see all of their skating in-between. I also wrote at the bottom the marks that each skater got for TR, SS, PE, IN, CO and STSQ and CHSQ GOEs.

It's unlisted because I don't want ISU to delete my channel, but you may reupload it or share it elsewhere.

(btw I'm sorry for the low quality)

[hoodie axel]

4 hours ago, WinForPooh said:

@ICeleste Bringing your video here!

 

The half-assed 'oh yeah must do a tr here' thing done four times is worth all that now.  

 

Yeah, friends and I have been discussing merits of the PCS as scored here (although we're firmly on "everyone's boring" train). Yuzuru is a more refined skater leading to more refined delivery of choreo, and I think his interpretation isn't bad, although not a fan of that or the choreo overall. Nathan had good moments in the choreo sequence and step sequence, and he has some interesting movement too, but a lot of his moves look throwaway at points because he simply isn't a good performer and I never find him cohesive... IDK what kind of point gap I'd have between them, but my initial impression of Hanyu>Chen stands.

 

(not bothering with Zhou)

[WinForPooh]

2 minutes ago, hoodie axel said:

Yeah, friends and I have been discussing merits of the PCS as scored here (although we're firmly on "everyone's boring" train). Yuzuru is a more refined skater leading to more refined delivery of choreo, and I think his interpretation isn't bad, although not a fan of that or the choreo overall. Nathan had good moments in the choreo sequence and step sequence, and he has some interesting movement too, but a lot of his moves look throwaway at points because he simply isn't a good performer and I never find him cohesive... IDK what kind of point gap I'd have between them, but my initial impression of Hanyu>Chen stands.

 

(not bothering with Zhou)

 

I was surprised because Vincent seemed a better performer to me there than Nathan. I didn't expect that because I've enjoyed Nathan's SPs and put my lack of enjoyment of his FS to just too many jumps, which means too much emptiness and setup. But I was very underwhelmed by his performance and interpretation here without the jumps, I think landing quads is a big part of the impact his skates have, maybe? And that leaves a far more positive impression of IN and PE when you watch his skates, which doesn't hold up when you ignore the jumps - only for components, of course. 

 

As for SS and TR... Yeah well okay.